Question $5$

Machine Learning

$[12$ Marks$]$

With regards to regression, describe how overfitting occurs and why it is problem$-$

atic.

Describe the basic idea behind $k-$fold cross validation.

How does a decision tree learning algorithm decide which attribute to test next?

$2$

Describe how the bandwidth parameter in kernel density estimation relates to over$-$

fitting and underfitting.

Describe how autoencoders can be pretrained for downstream classification tasks.

$2$